Hi Paul,

2010-07-18 01:13 Paul Libbrecht <[email protected]>:
> Isn't it pretty obvious that treating in OpenMath things outside  
> OpenMath is problematic?

It took me some time to realize that.  Before this discussion, I had thought
that the OpenMath CD language merely has an awkward way of dealing with URIs,
but otherwise is more expressive than RDFS.

> Don't we need some "functors" down here that would "create an OpenMath  
> something" based on some external description? Here's my 2p, using  a  
> functor external1#dlmf-description which would be interpreted as "a  
> symbol defined in the text pointed at the named URL with the classical  
> rigor of DLMF".

I think there is now a consensus emerging for representing these links outside
of the CDs.

> <FMP><OMOBJ>
>    <OMA><OMS cd="relation1" name="eq"/>
>          <OMS cd="transc1" name="sin"/>
>       <OMA>
>               <OMS cd="external1" name="dlmf-description"/>
>               <OMSTR>http://dlmf.nist.gov/4.14.E1</OMSTR>
>       </OMA>
> </OMA>
> </OMOBJ></FMP>

I don't think that a specific external1#dlmf-description is needed, as from
the http://dlmf... URI it's already clear that we are talking about the DLMF.
But what we would probably need, and what could solve the issue of translating 
csymbol/@definitionURLs to OpenMath, might be a generic String→URI
constructor, e.g.

<OMA>
        <OMS cd="external1" name="uri"/>
        <OMSTR>http://dlmf.nist.gov/4.14.E1</OMSTR>
</OMA>

> I would tend to feel this as being cleaner.
> And I have no preference between relation1#eq and owl#sameAs. I think  
> the more interesting question is what would be the equivalence of  
> subclass statements.

I didn't understand the last sentence, could you please explain it again?

Cheers,

Christoph

-- 
Christoph Lange, Jacobs Univ. Bremen, http://kwarc.info/clange, Skype duke4701

Attachment: signature.asc
Description: This is a digitally signed message part.

_______________________________________________
Om mailing list
[email protected]
http://openmath.org/mailman/listinfo/om

Reply via email to